Phase 2 and 3 clinical trials in Alzheimer disease (AD) and mild cognitive impairment (MCI) have been generally unsuccessful over the past decades. Efficacy mainly has been demonstrated in trials of 6 months or less, only with cholinesterase inhibitors for mild AD, and not in MCI or longer-term trials. Recent therapies based on the neuropathology of AD have not demonstrated efficacy in 18 month trials. Although this may reflect lack of knowledge of the biological targets and consequently the testing of ineffective drugs, it also reflects inadequacies of trials methods such that a modestly effective drug - if tested - may likely not be recognized as effective. In particular, the trials appear to choose patient samples based on assumptions about desired characteristics and outcomes that may not be valid because they are based on selected experiences from previous individual trials, further raising questions about design and validity. This project's broad, long-term objectives are to develop rational bases for designing pilot and phase 2 trials based on systematic evaluation of previous trials and testing design features prospectively. A broader goal is to encourage, complement, and accelerate the process of testing new and innovative treatments by providing models and simulations for clinical trials methods and designs. The research design and methods for achieving the objectives involves integrating patient data from 14 clinical trials and studies from the NIA's ADCS and ADNI over the last decade - including over 5200 followed from 1 to 4 years - into a common database. This allows for prospective analyses using hypothesis-based statistical techniques including simulations to better understand design parameters, outcomes, and to design better and more clinically relevant trials that would have the ability to detect change. The ADCS studies have inherent similarities that allow for combining, including common data base structures, outcomes, sites, training, and procedures. An expert steering committee formulates guiding principles and hypothesis-based research plans. Specific protocols address issues such as selection criteria, predictors of decline, and simulations to assess new trials designs. For example, we will examine the effects on outcomes of selecting patients for trials based on concomitant medication or severity using regression estimates and regression to the mean adjusted estimates, and follow this with trials simulations to empirically test validity and to plan future trials. Results of this research will have broad and lasting impact on trials designs in this therapeutic area, and especially on early phase trials in that designs can be tested and refined prior to implementing trials. The project is highly innovative in that it provides new approaches in this research area to examine the underlying patient data from numerous studies, and takes advantage of the rich database to empirically probe the assumptions under which current trials have been instigated. This will lead to improvement in trials methods and increase the likelihood for identifying effective treatments earlier in development.